A Synergistic ESO-MPC Framework for Robust Current Tracking in PEM Electrolyzers Under Nonlinear Voltage Disturbances

Proton Exchange Membrane (PEM) electrolyzers offer significant advantages for microgrid-integrated energy storage systems, including high hydrogen production efficiency and operational flexibility under variable renewable energy inputs. However, their highly nonlinear V-I characteristics introduce substantial current and voltage ripple when interfaced through power electronic converters, degrading power quality and accelerating electrolyzer degradation. To address this challenge, this paper proposes a synergistic extended state observer (ESO)-model predictive control (MPC) framework for power converters connected to PEM electrolyzers. In this framework, an ultra-local model is decomposed into a nominal linear component and a nonlinear disturbance term, where the nonlinear dynamics are estimated and predicted by an ESO. These estimated states dynamically update the MPC prediction model, compensating for nonlinearity-induced model uncertainties. Hardware validation using a physical PEM electrolyzer under real-world electrolysis conditions demonstrates that the proposed method achieves a 50.33% reduction in current ripple compared to conventional MPC approaches.